Natural History of Cutaneous Human Papillomavirus (HPV) Infection in Men: The HIM Study

Accumulating evidence suggests that cutaneous human papillomavirus (HPV) infection is associated with non-melanoma skin cancer (NMSC). Little is known about the natural history of cutaneous HPV. A sub-cohort of 209 men with no NMSC history, initially enrolled in the HPV infection in men (HIM) study, were followed for a median of 12.6 months. Epidemiological data were collected through self-administered questionnaires. Cutaneous HPV DNA was measured in normal skin swabs (SS) and eyebrow hairs (EB) for 25 and 16 HPV types in genera β and γ, respectively. Any β HPV infection was more prevalent in SS (67.3%) compared to EB (56.5%, p = 0.04). Incidence in SS was higher than 20 per 1,000 person-months for HPV types 4, 5, 23, 38 and 76. Median duration of persistence of β and γ HPV infection was 8.6 and 6.1 months in EB, respectively, and 11.3 months and 6.3 months, in SS, respectively. Older age (>44 years vs. 18-30 years) was significantly associated with prevalent (SS OR = 3.0, 95% CI = 1.2–7.0) and persistent β HPV infection (EB OR = 6.1, 95% CI = 2.6–14.1). History of blistering sunburn was associated with prevalent (OR = 2.8, 95% CI = 1.3–5.8) and persistent (OR = 2.3, 95% CI = 1.2–4.6) β HPV infection in SS. Cutaneous HPV is highly prevalent in men, with age and blistering sunburn being significant risk factors for cutaneous β HPV infection.


Introduction
Human papillomavirus (HPV) is a non-enveloped DNA virus, primarily infecting stratified epithelium [1,2]. HPV infections are broadly classified as mucosal (a genus) or cutaneous (a, b, c, m, Nv genera) [3]. An etiologic role for mucosal HPV infection has been established for several cancers, including penile [4] and cervical [5,6] as well as a subset of head and neck cancers [7]. In the majority of cervical cancers, viral DNA integrates into the host genome, further facilitating the transformation of infected cells [8,9]. Accumulating evidence suggests that, infection with cutaneous HPV types is associated with increased risk of nonmelanoma skin cancer (NMSC), in both immunocompetent and immunocompromised individuals [10][11][12], perhaps through an indirect mechanism. For example, inhibition of ultraviolet radiation induced apoptosis has been suggested as a potential mechanism of carcinogenic activity of cutaneous HPV [13,14].
The earliest evidence for oncogenic potential of b HPV types came from studies by Jablonska and colleagues, in 1972, which demonstrated the presence of b HPV types in cutaneous lesions of patients suffering from Epidermodysplasia Verruciformis (EV) [15], a rare hereditary skin disease that often progresses to squamous cell carcinoma (SCC) of skin in solar exposed areas. Cutaneous HPV DNA has been detected in eyebrow hairs, normal skin samples as well as tumor tissues from NMSC cases [16]. Several studies have reported a positive association between cutaneous HPV DNA and/or seropositivity and NMSC [17][18][19][20]. Despite its potential role in the development of SCC, very little is known about the natural history of cutaneous HPV infection. This information is essential to further elucidate the role of cutaneous HPV in skin cancer and also to guide future preventive measures targeted at reducing the burden of cutaneous HPV infection.
Few studies have evaluated the prevalence and persistence of cutaneous HPV among healthy individuals. Cutaneous HPV is thought to be ubiquitous in skin, with hair follicles serving as the reservoirs of persistent HPV infection [21,22]. High prevalence of cutaneous HPV in skin of infants and young children, indicates that exposure to HPV occurs very early in life [23], with asymptomatic infection persisting for several years [24]. Prevalence rates ranging from 42% in African population to 70% in European and Asian populations have been reported among healthy adults [25]. Prevalence of over 90% have been reported among immunocompromised patients [22].  (13.9) In this study, we describe the natural history of cutaneous HPV infection in a cohort of 209 healthy men in Tampa, Florida. We evaluated a spectrum of epidemiological measures including risk factors associated with prevalence, persistence and incidence across two genera of cutaneous HPV infection. We observed that cutaneous HPV is highly prevalent in men and that, age and blistering sunburn are significant risk factors for cutaneous b HPV infection in men.

Study population
The study population for the current analysis comprises a subcohort of men who participated in the U.S. site of the HPV infection in men (HIM) study, a large, multi-national prospective cohort study of the natural history of HPV infection in men [26,27]. The HIM study methods have been described in detail previously [26,27]. Briefly, between July 2005 and September 2009, study participants were recruited to the HIM study in Tampa, Florida, through mass advertisement targeted to university students, faculty, staff and members of the general population. Inclusion criteria were: 1) men aged 18-70 years; 2) reside in Florida; 3) have had no prior diagnosis of penile or anal cancers; 4) have never been diagnosed with genital and/or anal warts; 5) have not participated in an HPV vaccine study; 6) have no prior diagnosis of HIV/AIDS; 7) have no current penile discharge or burning during urination; 8) were not currently being treated for an STD; 9) have not been imprisoned or homeless during last 6 months; and 10) had not been in a drug or alcohol treatment program over the last 6 months at enrollment. Between November 2008 and June 2010, 1,082 participants in the HIM study were invited to participate in a supplemental study of the natural history of cutaneous HPV infection, requiring additional biospecimen collection. In the parent HIM study, the participants were followed every six months up to four years of follow-up. Nine hundred and sixty seven men enrolled in the parent HIM study had at least one sample of eyebrow hair or skin swab at their baseline visit. Of these 967 men, 965 men (99.8%) had three samples (one sample of eyebrow hair and one skin swab sample each, from sun exposed and unexposed skin), at their baseline visit. 89.4% and 66.7% men had all three samples at two and three visits, respectively. In order to maximize our observation time and facilitate estimation of incidence and persistence, the sub-study was restricted to those men who had all three samples for at least 4 visits (n = 211). Of these, 209 men had viable samples and were included in the final analysis.

Ethics statement
Written informed consent was obtained from all participants in the parent study, and a separate addendum to the consent was completed by all participants in the sub-study. The parent study and sub-study protocols were approved by institutional review boards at each recruiting site, including the Human Subjects Committees of the University of South Florida, the Ludwig Institute for Cancer Research, São Paulo, Brazil, the Centro de Referencia e Tratamento de Doencas Sexualmente Transmissiveis e AIDS, Brazil, and the National Institute of Public Health of Mexico, as described previously [27].

Data collection
Questionnaires. At the enrollment visit, HIM study participants completed a comprehensive self-administered questionnaire with information on demographics (age, race, education, and marital status), socioeconomic status, medical history, smoking status, alcohol consumption, and sexual history. Additional questions on risk factors for skin cancer (skin's reaction to season's first sun exposure, history of blistering sunburn, etc.) were added to the cutaneous HPV sub-study.
Eyebrow hairs and swabs of normal skin. Although, in the parent HIM study, swabs from both sun exposed and unexposed skin were collected, only swabs from sun-exposed skin were used for HPV DNA measurement in the present sub-study due to limited funding. A 565 cm area of normal, sun exposed skin, from the top of forearm, was pre-wetted (0.9% NaCl) and swabbed back and forth five times using a cotton-tipped Dacron swab (Digene, Gaithersburg, MD, USA). All swabs were placed in a separate vial and preserved in 500 ml Digene Standard Transport Medium (STM) for HPV DNA testing. Three to four hairs were plucked from each eyebrow (6-8 eyebrow hairs per individual) using disposable tweezers. Hairs with attached follicles  and c HPV testing, respectively, and n = 156 and n = 209 skin swabs from sun-exposed skin for b and c HPV testing, respectively. *Statistically significant difference based on McNemar's exact test (p,0.05). b c HPV types 101, 103, 108, 109 and 119 were measured in skin and eyebrow hairs but were not detected in our population at baseline. **Only for men who tested negative for any or type-specific HPV infection at baseline. 11 out of men who were positive for any or type-specific HPV infection at the first of two or more consecutive visits. c HPV types 101, 103, 109, 119 were not detected in skin and eyebrow hair. c HPV types 88 and 95 were not detected in eyebrow hairs but were detected in skin swabs of 2 and 1 incident infections, respectively. doi:10.1371/journal.pone.0104843.t002 were snap frozen in liquid nitrogen. All tissues were stored at 270uC until testing for cutaneous HPV DNA.
DNA extraction from skin swabs and eyebrow hair. DNA extraction was performed at the International     Briefly, the Dacron swabs were carefully cut into an Eppendorf tube using scissors and incubated overnight in proteinase K and buffer G2 (Qiagen, Hilden, Germany) at 56uC. An EZ1 DNA Forensic protocol was used to extract the DNA from eyebrow hairs according to the manufacturer's instructions. To monitor the possible occurrence of cross-contamination between the different specimens during DNA extraction, tubes containing buffer only were also included (one tube with buffer every ten specimens).
Viral DNA detection using Multiplex PCR/Luminex assay. The detection of viral DNA was performed by a multiplex PCR/Luminex assay [28], followed by the HPV typing using Luminex beads coupled with b or c HPV type-specific probes. The assay was performed using approximately 100 ng of total DNA and specific primers amplifying a part of the E7 gene for 25 HPV types in genus b (5,8,9,12,14,15,17,19,20 [28,29]. Although, a large number of HPV types are being continually discovered, a limited coverage of HPV types was offered in the multiplex assay. Two primers for the amplification of b-globin were included in the assay to provide a positive control for the quality of the template DNA. The positivity of the assay was given by the intensity of the fluorescent signal detected by the Luminex apparatus and was expressed as the median fluorescence intensity (MFI) of at least 100 beads per bead set. The cutoff was calculated for each HPV-specific probe by considering the MFI values obtained with no respective PCR product. The cutoff was computed by adding 5 MFI to 1.16 the median background value, as described by Schmitt et al. [30]. All MFI values above the cut-off have been considered positive. PCR was performed with the QIAGEN multiplex PCR kit according to the instructions of the manufacturer. This multiplex PCR protocol is highly sensitive, being able to detect only 10 copies of the viral genome [28,29]. HPV genotyping was successfully repeated in a blind manner, three times in 10 individual subjects, demonstrating reliability of multiplex PCR for detection of specific HPV types [28]. Similar strategy was used for determining the sensitivity and specificity of c HPV assay, showing features similar to those of the b HPV assay. Following PCR amplification, 10 mL of each reaction was analyzed by multiplex genotyping using a Luminex based assay as described [31]. Results were expressed as the median fluorescence intensity (MFI) of at least 100 beads per bead set. Of the 1,385 samples, 1,234 (89%) were b-globin positive. Thirty (2%) b-globin negative and HPV positive samples were included, while 120 (9%) b-globin negative and HPV negative samples were excluded from the analyses. All the tubes containing buffer only were tested negative for HPV DNA and b globin.

Statistical analysis
Baseline characteristics, including demographics, skin cancer risk factors, lifestyle factors and history of sexually transmitted diseases, were summarized for the cohort using descriptive statistics. HPV prevalence in eyebrow hairs and normal skin swabs was estimated by phylogenetic species along with concurrent infections with other HPV types. Type-specific prevalence of HPV was defined as the proportion of men who were positive for a given HPV out of the baseline cohort with valid samples. Genusand species-specific prevalence was defined as the proportion of men who were positive for at least one HPV type in the given genus or species. The statistical significance of differences in the prevalence of HPV infection between eyebrow hairs and normal skin swabs was tested using Fisher's exact test. Incidence (in person-months) was determined for any HPV infection, as well as type-specific and species-specific HPV infections in normal skin swabs or eyebrow hairs. For analyses of incidence of any HPV infection, the participant had to be negative for all HPV types (genus b or c) at baseline, and only the first incidence of any HPV infection was taken into account. Similarly, for species-specific and type-specific HPV infections, the participant had to be negative at baseline for that particular HPV species or type, respectively. Time to incidence of HPV infection was defined as interval, in months, between baseline visit and first visit with a positive HPV sample or time point being censored. Due to fewer events, median time to incidence of c HPV infection in eyebrow hairs could not be estimated. Time to clearance of HPV infection was defined as interval (in months) between incident HPV infection and clearance of all type-specific HPV infections. Persistence of any HPV (genus b or c) or type-specific HPV was defined as any or type-specific HPV infection at $2 consecutive visits. For a given type of HPV persistence, only participants who were positive for the HPV genus or type and had at least one follow-up visit after that were included. Kappa coefficient (k) was used to determine the concordance of viral infections across eyebrow hairs and skin swabs for each b-HPV type. The kappa coefficients ,0 indicates no agreement, 0-0.20 as slight, 0.21-0.40 as fair, 0.41-0.60 as moderate, 0.61-0.80 as substantial, and 0.81-1 as almost perfect agreement [32].
Logistic regression models were used to estimate age-adjusted odds ratios (OR) and 95% confidence intervals (CI) for the associations between risk factors and prevalence of overall and species-specific HPV infections (b1 and b2) (due to sample size constraints, risk factors associated with type-specific HPV infections were not examined.). Variables that were significantly associated with b HPV prevalence in an age-adjusted model were further evaluated after stratification by b HPV species. Logistic regression was conducted to estimate the association between risk factors and incidence as well as persistence of HPV infection, after adjusting for age. Logistic regression analyses were conducted separately for HPV infection in eyebrow hairs and normal skin swabs. Kaplan-Meier curves were derived for incidence and clearance of any HPV infection. An alpha level of 0.05 was considered statistically significant. Adjustment for multiple comparisons was not conducted. All analyses were conducted using SAS software, version 9.3 (SAS Institute Inc., Cary, NC, USA) and R software, version 2.13.1.

Results
The median follow up time was 12.6 months. As seen in Table 1, majority (74.5%) of men self-reported their race as White and was between 18 and 30 years of age (51.7%).

HPV DNA in eyebrow hair and normal skin
As seen in Table 2, cutaneous HPV prevalence was higher among normal skin swab samples than eyebrow hairs, particularly for any b HPV type (67.3% in normal skin swabs vs. 56.5% in Table 4. Association between baseline characteristics and prevalence of species-specific HPV infection in men residing in Tampa, Florida.    As seen in Figure 1, the median time to incidence of cutaneous b-HPV infection was 17.3 months in eyebrow hairs with a slightly shorter time to incidence in normal skin swabs (11.2 months). Thus, at 12 months, 35% of men had newly acquired cutaneous b-HPV infection in eyebrow hairs while 57% men acquired new cutaneous b-HPV infection in skin swabs. The median time to clearance for any b HPV infection was 6.1 months in eyebrow hairs and 6.44 months in normal skin swabs (Figure 2). For c HPV infection, the median time to incidence in normal skin swabs was 13.1 months (Figure 3), while the median time to clearance in both, eyebrow hairs and normal skin swabs, was 6.1 months and 6.4 months, respectively (Figure 4).
Incidence and persistence of cutaneous HPV are presented in Table 2. In general, incidence and persistence rates were higher in normal skin swabs compared to those in eyebrow hairs for b and c HPV infection. The top five HPV types with highest incidence of infection in eyebrow hairs were 38 (b2), 5 (b1), 4 (c), 80 (b2) and 23 (b2), ranging from 5.7 to 7.6 cases per 1000 person months; while in normal skin swabs these included 4 (c), 23 (b2), 38 (b2), 5 (b1) and 76(b3), ranging from 21.7 to 28.5 cases per 1000 person months. More than 50% of men had persistence of HPV types 25, 47, 12, 15 and 17 in eyebrow hairs and persistence of HPV types 20, 21, 47, 23, 50, 108, 25, 37, 76 and 8 in skin swabs. Median duration of persistence of b and c HPV infection was 8.6 and 6.1 months in eyebrow hairs, respectively and 11.3 months and 6.3 months, in skin swabs, respectively.

Risk factors associated with prevalence, incidence and persistence of HPV infection
Men aged .44 years were more than twice and thrice as likely to have prevalent baseline b HPV infection in eyebrow hairs and skin swabs, respectively, compared to men aged 18-30 years (Table 3). A positive history of blistering sunburn was associated with a more than two-fold increased prevalence of b HPV infection in normal skin swabs, although the association between frequency of sunburns and HPV prevalence did not reach statistical significance (Table 3). No other demographic, lifestyle or sexual history related factors were associated with prevalence of HPV.
Evaluation of species-specific b HPV prevalence indicated that men aged .44 years were twice more likely to harbor b1 and b2 types of HPV in eyebrow hairs and more than threefold likely to have b1 and b2 types of HPV infection in normal skin swabs (Table 4), compared to men ages 18-30 years. Similarly, a positive history of blistering skin burns was associated with more than two fold (OR = 2.6, 95% CI = 1.2-5.7) and more than threefold (OR = 3.1, 95% CI = 1.4-7.0) increased prevalence of b1 and b2 HPV in normal skin swabs, respectively. This association was stronger for b2 HPV infection in normal skin swabs (OR = 5.0, 95% CI = 1.4-17.7) among individuals with a history of more than 2 lifetime episodes of sunburns compared to those who reported none. Overall, while the association of these factors with b HPV infection did not vary significantly by HPV species, age and history of blistering sunburns appeared to have strong positive associations with b HPV infection, particularly in normal skin swabs.
No significant associations were seen between baseline factors and incident b HPV infection (Table 5) and prevalence or    Had an alcoholic beverage in the past month  (Table 6). Findings on factors associated with persistence of b HPV infection were similar to those of prevalent infections described above. Age.44 years was associated with a six fold (OR = 6.1, 95%CI = 2.6-14.2) increased persistence of infection with b HPV in eyebrow hairs (Table 5) but not in normal skin swabs. History of blistering skin burn was associated with persistence of b HPV in normal skin swabs (OR = 2.3, 95% CI = 1.2-4.6), with increasing frequency of sunburns showing stronger association (Table 5). When restricted to White men, history of blistering sunburns was associated with 2.7 times increased persistence of any HPV infection in normal skin swabs (OR = 2.73, 95% CI = 1.19-6.26), after adjusting for age (data not shown). As expected, this association was stronger than that observed for the overall population (OR = 2.3, 95% CI = 1.2-4.6) ( Table 5). However, when analyses were restricted to White men and stratified by HPV species, there were no significant associations between history of blistering sunburns and persistence of b1 (OR = 0.87, 95% CI = 0.29-2.62) or b2 (OR = 1.33, 95% CI = 0.54-3.27) infection in normal skin swabs, after adjusting for age (data not shown).
No significant associations were observed between baseline characteristics and persistence of c HPV except a significantly increased persistence (OR = 4.0, 95% CI = 1.0-15.9) in eyebrow hairs among former smokers compared to never smokers ( Table 6).

Discussion
In this HIM study sub-cohort of 209 healthy men, we observed a high prevalence of cutaneous b HPV and c HPV in normal skin swabs, higher than that in eyebrow hair samples. Although cutaneous HPV seroprevalence rates have been reported previ-ously [33][34][35], seroreactivity is not a direct estimate of HPV infection. Hence, we attempted to directly measure type and species-specific HPV DNA in normal skin swabs and eyebrow hairs.
Results presented here contrast to that of other studies which reported a higher prevalence of cutaneous HPV in eyebrow hairs among individuals without a history of skin cancer [36,37] and lower prevalence (13-37%) in skin [37,38]. Tarmorshuizen et al. reported a prevalence of 54% in eyebrow hairs among healthy controls [39], which was similar to our findings. Differences across studies could be a reflection of variation in number of HPV types examined. For example, while we evaluated 41 HPV types (25 HPV types from genus b and 16 HPV types from genus c), previous studies have evaluated fewer HPV types within each genus (6 to 28 HPV types overall across studies) [36][37][38][39]. Apart from this, methodological differences or even differences in geographical location of study populations may explain the inconsistencies in HPV prevalence across studies. Indeed, HPV prevalence rate has been shown to vary by country [25,36]. In our study, while we observed a strong association between history of blistering sunburn and both prevalence and persistence of b HPV, no association was seen between tanning ability and b HPV infection, indicating that exposure to ultraviolet (UV) radiation rather than host genetics may be more important determinants of cutaneous HPV infection. Thus, geographical variation in UV radiation might have contributed to variable prevalence estimates in the literature. While we were unable to measure HPV prevalence in sun-unexposed skin, our previous study of NMSC cases indicates that the prevalence of b-HPV is higher in sunexposed skin swabs (100% prevalence) compared to sununexposed swabs (95% prevalence) among NMSC cases [16].
Older age and history of blistering sunburns were significantly associated with both prevalence and persistence of HPV but not Ever been diagnosed with an STD Natural History of Cutaneous Human Papillomavirus (HPV) Infection with incidence of cutaneous HPV infection. Our findings are consistent with previous reports on the association between age and prevalence of HPV [36,39]. A positive history of blistering sunburn was associated with a more than two fold increased prevalence of b HPV infection in skin, with a stronger association for b2 HPV infection. While an inverse association between frequency of sunburns and b HPV seropositivity has been reported previously [33], we found a positive association between lifetime history of one blistering sunburn and prevalent b HPV infection in normal skin swabs. We also observed a strong association between frequency of blistering sunburns and persistence of b HPV infection, likely explaining the elevated prevalence observed here. UV exposure mediated immunosuppression may predispose individuals to a higher risk of cutaneous HPV persistence and hence prevalence [40]. Indeed, higher prevalence of cutaneous HPV has been observed among individuals who reported working outdoors for longer duration [41].
The findings of our study should be interpreted with caution. The external validity of these findings is limited to men. Future studies should include women, especially given the previously reported differences in seroprevalence of specific types of cutaneous HPV by gender [35]. Our findings are generalizable to non-Hispanic Whites only. However, our study population of Florida residents is at higher risk of NMSC compared to other U.S. states, corresponding to Florida's higher UV radiation index [42], and hence may represent a population that could be targeted for novel NMSC prevention measures. To maximize the sample size, persistence of HPV infection was determined using both incident and prevalent HPV infections. Therefore, the true duration of HPV infection is unknown, given the prevalent cases at baseline. Finally, we examined only 25 HPV types in genus b and 16 HPV types in genus c. Given the large number of HPV types that are being continually discovered, our results on overall b and c HPV infection may not reflect the true associations that can be affected by unexamined and unknown HPV types.
The study has several strengths. This is the first study to report HPV infection across a continuum of incidence, prevalence, persistence and clearance of cutaneous HPV across different genera, by type and species. We compared HPV infection in normal skin swabs as well as eyebrow hairs, thus providing a comprehensive report on natural history of HPV using biomarkers from two tissues. This is particularly important since concordance between HPV infection in normal skin swabs and eyebrow hairs was found to vary by HPV type. Along with rates of HPV infection, we also evaluated a range of epidemiological factors associated with HPV infection, overall and by species. Analyses were adjusted for potential confounders.
In conclusion, we observed high prevalence, incidence and persistence of cutaneous HPV infection among cancer-free men. Age and history of blistering sunburns were significantly associated with prevalence and persistence of b HPV. The association of blistering sunburns with both persistence of HPV infection, but not with incidence, suggests that sunlight exposure does not affect acquisition of new HPV infection but the duration of HPV infection could be affected by sunlight exposure through UV related immune dysregulation or actual promotion of HPV infection. Given the potential role of cutaneous b HPV in NMSC, the findings are valuable in defining a high risk population for development of novel preventive measures.